Daisy, an AI granny wasting scammers' time
news.virginmediao2.co.uk
news.virginmediao2.co.uk
A couple years ago I would sit at my desk thinking about a really hard problem in silence. The phone rings. Spam call. Every 30-180 minutes another one. If you now think turn the phone off, well not that easy as CEO of a business when people expect you to be reachable.
It creamed my corn so much that I recorded my own voice samples as a senile "Opa Denny" (german grandpa Denny), modelled after Lenny. Complete with background ducks hanging out on the couch to Opas dismay, later in the call. It works on autopilot without interaction because on Asterisk, and with the largest German SIP provider at least, you can extract the calling peer identity from the SIP header. So I wrote a scoring system based on indicated number, black and whitelist regexs for number and for calling peer, greylist for the geographically surrounding number prefixes, etc. A legit mobile call would show up as number@t-mobile.de for example, while a spam call would say fakenumber@01012.com.
Asterisk would record the call in wideband stereo, normalize the audio, and mail it to me as MP3 attachment. Funny for a while, but these days I just throw all such calls onto the mailbox. Since they need a real person to scam or create a sale, the call is finished right away.
It works great to this day, because I never published it.
Voicemail
Spam calls (and even more so: SMS) have gotten much more frequent on my mobile number in the last 2 years. But in this case, it's a number that has been in active use for the last decade and has by now probably been in multiple data leaks... :/
Come again?
Edit: I guess it is in urban dictionary, but my first thought was the last definition listed:
https://www.urbandictionary.com/define.php?term=creams%20my%...
I never get spam calls or sellers calling me.
When a big telecom does it, the second thing they do with it is to fuck up the spam detection so bad that every third phone call I make gets answered by "Daisy".
And just think about it - why would a telecom need this tech? They can already drop the spam calls and stop routing calls from the bad actor telecoms who enable the spammers. They don't do that because they prefer to collect a few cents a call from them rather than serve their customers better. It's everyone else who needs this.
Not connecting calls doesn't waste spammer money, but maybe Daisy does.
If the big telco can find 10 righteous callers from a a bad actor telecom, they should keep routing the calls.
Voicemail that is left in this generic voice mail box never makes it to their customer and the customer is completely unaware that some of their calls have been diverted.
This is the thing we screwed up for email and phone (after per call fees dropped to zero).
It's not rocket science to create systems that net to zero for common usage (balanced in-bound vs out-bound), but charge an arm and a leg for bulk senders.
If a scammer puts "FINAL NOTICE" on a solicitation they mailed with no prior relationship, I do still report it as fraud. But that's probably wishful thinking.
I doubt very seriously that the pool of people who have knowingly and intentionally and explicitly opted in/consented to telemarketing - that is, without any dark pattern involvement and with a clear and unmbiguous consent experience, is very large. In fact I think it is infinitesimal because I can’t recall seeing such a consent UX- they ALL involve dark patterns. And if you pair that with “marketer who diligently implements all state & FTC requirements and does timely and accurate processing of removal requests, I think that the 3 relationships left are web app UX testers.
I think the world would be a better place without telemarketing or email marketing. Maybe a “one email per year” limit per merchant who you have actually paid money to and not opted out of.
If the telecoms can perfectly predict the telemarketers, then I’d love it. But in practice how often is this going to block people I know from calling me? Probably not never, and then we just have to give up on phones as a reliable method of communication.
The disadvantage of email is that it doesn't work for urgent use cases, like Disneyland calling the parent of a lost child.
(Hopefully leaving voicemail keeps out of reach of spammers, I guess it's just too expensive for them ?)
I don't think any telco puts effort into stopping spammers.. I'd like them to but I don't think it's something they either can care or legally capable of fixing.
Thank you for the insight.
This is actually genius. The spammers will have blacklisted all of their targets eventually.
presumably someone is still getting charged for calls and AI -- and therefore someone is making money -- in this situation though, non?
Why?
> Influencer and reality TV star, Amy Hart, has worked with Daisy to produce a shocking video to show how she’s taking on phoney fraudsters
I'm not saying this fixes everything, but I would rather a world where scammers odds at making a living at this are so poor they won't bother versus a world where everyone has to block every number by default and live in metaphorical bunkers to never interact because you might be a scammer.
Then what's stopping the scammers from finding another "evil" job that makes money? You have to remember that humans tend to not enjoy accomplishing thievery and the scammers most often do this out of necessity. Of course there are big call center operators who truly are terrible people, but this conclusion is true for the bulk of their workforce.
There are endless ways to earn an honest living. Nobody stealing from vulnerable people and the mentally degraded is forced to. I fully believe that people can survive without scamming seniors for a paycheck.
Personally, my default was to pick up and is now if recognized contact pickup, if not and I'm expecting an unknown call, scrutinize, then if pickup, only answer with a cough or two — never "hello" or "yes" due to threat of voice cloning escalate to banks.
Spam is universally detested
My assessment is that voice calls are on the verge of going obsolete if the telcos fail to get a handle on spam. Yet the telcos behave as if they have no clue whatsoever and DGAF.
Consider: if you get harassed with a call or text you don't like, you send an SMS with that phone number to some known short code. You then receive $1 from the caller. If the caller cannot be found, the last verifiable link in the chain is responsible for paying the fine.
This would cause carriers to behave overnight, instead of allowing foreign call centers to spoof other people's real private numbers with your area code.
plus a lot of spam comes out of compromised phones, or compromised systems. what good is this if i have your phone, or just your SIM?
Where you are expecting a call and it's definitely not spam, but they're not going to be ringing from a pre-approved whitelist. Don't think they'd want to be putting $1 per call on the line and hoping for people to remember to click the right button afterwards
There are probably downsides and ways this will screw up real relationships but it will certainly increase the cost of spam.
- Every contractor (plumber etc) you hire will ask you to please add them to your contact list first so that they can message you.
- After a while of half their clients not doing that and lots of fees on their end, contractors stop providing a phone number at all, asking you to please install ContractorApp to communicate with them.
This message is from an unknown number. (Accept / Block / Charge sender $1)
It’s a boy!
That said - perhaps harder to change the name that shows up as quickly as you could leave a recorded name? :)
Still one of the best ads ever made.
I disagree about the we-adda-baby-itsa-boy issue. I don't see how that'd apply given that you can charge them $1 from the very first message.
contractors can add this to their invoices if they care.
>Every contractor (plumber etc) you hire will ask you to please add them to your contact list first so that they can message you.
This is reasonable. If they want to reach me, they should whitelist in advance instead of hoping they can randomly get through.
Best I can offer you is an in-game bonus for watching an ad.
Like if you are interested in the convo then you should be paying the other person as you are getting value from the convo.
What's the situation where you don't care about the convo but for the right price you could care, and the other party also thinks that price is reasonable? Like maybe if someone is trying to recruit you i guess, but the situations where that is true seem very few and far between (and we already have a system for that, where traditionally someone offers to buy you a meal in exchange for listening)
Marketers seem to think so, or they wouldn't waste their time calling.
Minimally targeted advertising.
Like if you assume a CPM of $2, that means $0.002 per phone call. Would you really find that acceptable?
If the answer is no, then you aren't really accepting payment,you are just blocking people with extra steps. No different than if you said you will accept a phone call for a billion dollars. You aren't really accepting something for payment if you set a price that nobody will pay.
(Sorry if I'm being blunt, attention economy participants can't be so they'd sugar coat it to such a degree that you might miss the point).
https://mmalmi.github.io/satoshi/
> "[this next bit turned out to be very controversial. there is extreme prejudice against spam solutions, especially proof-of-work.]
> It can already be used for pay-to-send e-mail. The send dialog is resizeable and you can enter as long of a message as you like. It's sent directly when it connects. The recipient doubleclicks on the transaction to see the full message. If someone famous is getting more e-mail than they can read, but would still like to have a way for fans to contact them, they could set up Bitcoin and give out the IP address on their website. "Send X bitcoins to my priority hotline at this IP and I'll read the message personally."
Every player can configure an amount of ingame money that is levied from a sender's account to deliver a message to them. It's a currency sink, so it's themed as a "tax" levied by the NPCs and its value is destroyed rather than paid to the recipient of the message.
I thought it used to default to 5 ISK (a pittance, something you can make back by shooting a single NPC pirate ship). I see some references to the default being ~2000 ISK at the time that it was changed to 0, where it is now.
Worked pretty well, imo. Players that need to be publicly contactable (people who organize public events, for instance) can turn it off easily. People who are "space famous" can crank it up to reduce targeted spam. Even at the default setting, it's effective at keeping ingame scammers from blasting the whole player list with messages (at least, the poor ones :). Doesn't apply to people you've already exchanged messages with. I think there's also some allowlisting you can do, etc.
0. https://wiki.eveuniversity.org/EVE_Mail#CONCORD_Spam_Prevent...
The idea in general seems to have been around since 1992: https://en.wikipedia.org/wiki/Hashcash
It’s a cool idea and would work good for real world fines, too: fines that destroy the money paid instead of transfer it (to the city, the court, the police and so on). The fines would disincentivize bad behavior, while also not incentivizing the police to “go generate some crime” because it pays them.
MMT (modern monetary theory) argues that taxation isn't about "funding", it's actually about controlling inflation.
MMT is somewhat controversial.
But if those taxes go to the State of California, it's different. They don't print money, so it becomes a real debt to them. Destroying it in that case would cancel that debt and free the debtor (the Treasury) from that obligation.
You can buy jigs that hold hundreds of SIMs connected to a handful of cellular modems connected to a c&c and drip them out.
For me, this has been surprisingly effective against robocalls. Obviously this isn’t going to work against scammers who call directly, but most of the spam calls I receive start with some pre-recorded message which isn’t going to pass the menu.
Edit: s/auto/pre
It sounds surprisingly human-like, even saying "Hello?" in a slightly annoyed tone when the other person doesn't respond in time.
Another good one is for the phone to stay on hold for you. That one has been extremely valuable to me as Qantas would regularly keep me on hold for over 5 hours when I tried to get my money back for a cancelled flight. The operator would sometimes be a bit confused when I pick up, but it usually worked well and certainly beats listening to hold music for hours.
I remember being an outside sales rep for a local mom and pop wireless company in the late 90's, early aughts. We sold an automated "assistant" called Wildfire that would screen calls and stood as an intermediary between you can callers. She would answer, you would record your name and then it would call you on any number of devices you had entered. At one point, I had it calling three of my numbers (office and two mobile numbers) and at any time, you could just send people to voicemail. It was very similar to how many of the AI assistants work today. If I remember it was like $30/month, but as reps we got to use it for free which was really fun tinkering with it.
AT&T also had something like this where you could have a program they offered which would screen your calls and either connect you or you could send people directly to voicemail. It didn't have nearly the features that Wildfire had, but it was effective.
Obviously in the late 90's and early aughts, something like this wasn't really needed and after a few quarters, AT&T quietly stopped offering their service. Wildfire lived on until the mid aughts after being bought and then killed due to lack of adoption and use.
Kind of crazy these kinds of programs were pretty common before the AI assistant craze now.
Details about Wildfire here: https://en.wikipedia.org/wiki/Wildfire_Communications
Scammers are a stain on the reputation of India. You could argue it's unfair to tar an entire country with the same brush, but quite clearly rule of law isn't properly functioning over there and there's complicity in letting them do this. Same goes for Nigeria.
You can get away with murder in India if you're wealthy enough (e.g. the case of Jessica Lal, murdered by a politician's son in front of at least a dozen witnesses). The egregious corruption of the INC or "Congress party" (which is ideologically progressive) over many decades has created a massive voter exodus to the conservative BJP party, the majority party in India since 2014. However, the corruption and inefficiency at all levels of civil society has remained endemic.
Jim Browning (the scambaiter who worked with O2 in this article) has successfully compromised several scam operations, gotten their physical address and other dox, and referred them to the police. The offices get raided, Jim gets some nice CCTV footage of the raid, the operators of the criminal enterprise get a nice perp walk... and then a month later the case is mysteriously dismissed with a bullshit reason about AI deepfakes and "IO" (influence operations, I presume).
The thing to keep in mind is that India's government is run by Modi, a Hindu ultranationalist who wants to deport all the country's Muslims to Pakistan[0]. There's a pretty straightforward pipeline from organized crime to fascism and I wouldn't be surprised if the scammers in question here are part of Modi's power base (or part of other organizations which are part of his power base).
The only thing I could think of to fix this would be to strategically suck people out of India through generous visas for migrants who want to live in a country with functioning[1] institutions. The thing about organized crime is that it relies on having a pool of suckers to continue joining the criminal enterprise - in other words, even the scammers are themselves being scammed. This is one of the less selfish reasons why I'm an open borders fanatic, but I also have to admit that such a policy in today's era has negative political capital.
[0] Which itself has money problems because it's budget gets siphoned off by their own military and they have to beg the IMF for scraps
[1] To be clear, India's institutions still exist, they're just mildly broken.
It's like corporate avatars for hiring or cold call bots for sales.
I'd say stick to white listed numbers, but pro phones can't do that, and they are the most prone to spam.
It finally worked the other day! A robot repeated "chrysanthemum" back to me when asked. Sadly it seemed to be limited to this one single trick; as I continued my attempts to exploit it, it abruptly apologized and ended the call.
Cute, but one day I am going to get lucky and a scammer is going to get absolutely wrecked.
Doubling the dwell time for a scammer will halve their profits. That could have interesting second-order effects. Perhaps it makes it not worth it for some subset?
So I don’t think that this is just entertaining PR, I can see why it’s better than simply banning the scammers. Still a question of cost though.
A person who is shadow banned generally continues their antisocial behavior in the face of multiple warnings and reprimands. I don't see it as cruel. I see it as defending the community against those who get pleasure out of trying to wreck it. It's effective because the bannee is generally too dim to realize what has happened to them until they get bored and go away.
It's very easy to check if your posts are available by browsing them anonymously or under a second account.
There are some other Youtube videos where a guy writes these ChatGPT personalities, but those aren't as entertaining because the AI is basically just blurting out random thoughts rather than engaging in an actual conversation.
I guess it pretty much means retiring this number next year. I’ve inly had the number a little over a year and the spam has slowly crept up despite best efforts to give fake numbers where possible.
Might put the number of someone I don’t like first; then flip the privacy setting and fix the number
There is one simple and easy solution to this mess, but good luck getting traction on it.
It'll help with so many things: - in contact syncing systems you can't rainbow table your way to decrypting numbers - numbers can be permanently burned once they're released or deemed as spam. This means every service could ban spammers safely without fear of burning a real user. - people could more easily have alt numbers, non-voip numbers for untrusted services.
Your suggestion won't help circumvent that. I think.
> and one in five (22%) experiencing a fraud attempt every single week
That seems low. I guess they mean phone calls only.
I get -no exaggeration- several hundred spam mails a day (and these are the ones that made it past the first line defense). I also get 10 or more scam texts.
Some of the phising emails that I get, are frighteningly realistic (but invariably seem to have at least one speling eror in them).
If folks live in a normal suburban development, they are highly unlikely to get several hundred crooks, knocking on their door, every day, but being on the Internet, means exactly that.
The crooks only need to land you once, you have to have a perfect record, avoiding them.
I've seen no legit spam e-mail in years, on my very old, very used, "forums and other logins I don't care about", 2005 gmail address. What provider are you using?
I have an OG address that was registered about five minutes after Steve introduced the service.
Network Solutions sold it to spammers, and the rest is history.
Also, a couple other of my emails (and phone numbers) are associated with businesses, so are sort of public record (not really, but everyone finds out, anyway).
It's long past time that phone calls should stop being synchronous and immediate. They are only this way because historic landlines were synchronous and immediate, and nobody changed it. This functionality should be culturally unacceptable in 2024.
Imagine if the concept of a land line never existed. Then, a startup came out with an app that: 1. allowed random people to cause your handheld device to stop what it was doing and immediately ring and vibrate and pop up a dialog over whatever app was active, 2. when you pushed a button, an instant two way audio channel was initiated, and 3. they could do all this without any authentication whatsoever--just needing to know a non-secret 10 digit number. This kind of app would fail every App Store guideline, and it would be laughed out of the ecosystem as unacceptable to end users. But in reality, it's "The Phone App" that comes pre-installed on your damn phone!! Insane that this kind of application is acceptable.
Joke aside, is this going to be cost effective? What would it cost to keep a scammer on the phone for an hour? Who will pay this bill?
Robogranny Olympics. Allow people to submit their own bots, Micromouse/koth.org/Battlecode-style. Bots are run on random phone numbers for several weeks, and are ranked on average call length (total time/num calls). The robot that can keep spammers on the line the longest wins the prize.
A condition of submission is that the bot code and weights are published (not open sourced, just published). Competition is repeated yearly.
https://youtu.be/V2IyttWHJfs?si=mreFXk6h-yDAJ3D7&t=50
I've had a few friends get new numbers, and seeing "Hey it's your Aunt, I got a new phone" live transcribed on the screen has come in handy.
They're unlikely to do this, because scam call centers are typically a branch of local organized crime syndicates. There is nowhere for them to roll, because their boss' boss also pays a bunch of crooked cops and whatnot.
The bottom rung guys doing the actual calling aren't gangsters, but further up the organization they are, so attempting to fuck with the organization from the bottom will not end well for you if you live in that area and lack protection.
Strangely, I looked at OpenAI's offering.
Despite it being crazy expensive, they also didn't offer a reference implementation anywhere close to the functionality that is demo'ed in the regular ChatGPT app.
Nobody seems to get the interruptions correct - people are doing all sorts of weak workarounds like push-to-talk, etc.
This is called a totem.
You’ve invented or learned a caricature to rail at which may have once been based in truth, and from time to time again approximates it, though never with the fidelity you ascribe to the original. It’s commonly done by sides in partisan polarisation, the most common being a two-mode system that pillories its picture of the other.
If you picture the person writing the totem comment, you probably have a clear idea of what they do for a living, how they dress, et cetera. Totems are why both deification and demonisation work; they’re a hack of the human ability to visualise and project.
Analogy: phantom pain is pain in a limb that was once there but now isn’t. The limb was real. It probably felt pain. But when a patient imagines pain after the limb is gone, that’s the diagnosis.
Those comments are real. The people, probably, too. Responding to them when they aren’t in the room is a separate matter.
This is the totem. You react to a misguided belief even when it’s a phantom.
If you look at two-mode polarised discussions, you’ll find both sides talking about a totem of the other, reacting to imagined preëmptions and rarely interacting with each other. The graphs separate. As a result, the preëmptions are more imagined than real. (Both in characterisation and frequency of emergence.
Note that this is perfectly normal. Kids do it. And it’s fun. It’s also easy, since instead of reacting to anything empirical or well argued you’re constructing straw men for the purpose of taking them down.
I clarified my position. The visualisation isn't the totem per se. The representation is.
> term carries a negative connotation, and therefore proves the other person (myself) correct
Nobody said you were wrong. Correct, necessary and even germane are distinct.
The above comments seem to me much like a simulacrum of how that might go. The comment exchange should have stopped much sooner.